On 2010.10.12 09:34:10 +0100, Chris Wilson wrote: > On Tue, 12 Oct 2010 16:13:51 +0800, Zhenyu Wang <[email protected]> > wrote: > > This adds new ring for blitter engine starting from Sandybridge. > > Thanks, look fairly straightforward. The messy parts are a clear reminder > that I intended to clean up some areas for multiple rings. > > * bikeshedding > > s/HAS_BLIT_SPLIT/HAS_BLT/ -- the documentation has ingrained BLT engine > into my head (and I hope they haven't changed on a whim for snb)
ok, BLT looks better. > > After initialisation we should not need to care about HAS_BSD/HAS_BLT and > just test whether the ring is initialised or active (depending upon op). > That should dramatically reduce the number of tests in our code and > hopefully lead to further simplifications. > > Unified request list (along with per-ring lists) will sort out the mess in > i915_gem_evict and also solve the wrap-around issue. > > So no get/put irq again? yeah, current still uses user_interrupt, that has been unmasked and enabled from beginning. I just did some copy&paste, it could be fixed to our original behavior by mask initially and unmask as required, which might involve irq enable/disable sequence specific on different hw though. -- Open Source Technology Center, Intel ltd. $gpg --keyserver wwwkeys.pgp.net --recv-keys 4D781827
signature.asc
Description: Digital signature
_______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
